dhcpcd-discuss

On Sat, Sep 13, 2014 at 11:53:46PM +0100, Roy Marples wrote:

[snip]

> 
> OK, the timeout you are seeing looks correct from the 1st dhcpcd instance 
> spawned. You're spawning the 2nd instance which says the 1st one is running 
> and then the 2nd one exits - it does not affect the 1st instance in any way or 
> form.
> 
> dhcpcd has it's own carrier management, you don't need to spawn a 2nd copy on 
> carrier up. If you do want to do this, ensure one isn't already running!
 
Thanks for replying.

The interesting thing is I compiled the latest version from tip and it
works as expected. I will get you the logs when I get back on Monday. It
seems as if in version 6.4.3, when dhcpcd sets an 8s timeout for IPv4LL,
and then a dhcp offer is received, it still exits.

Maybe the following commit fixes it in tip:

eac782e8d2cbfbd67d76920ba9e9ebcf22aecf61
Fix expire and re-solicit on drop

With dhcpcd's carrier management, the daemon itself is spawned for eth0
and then times out and exits (30s I think). So you're saying to set
dhcpcd timeout to 0 so that the daemon never exits and thus handles
carrier detection?


Thanks,
Amit